Accelerated prediction of lattice thermal conductivity of Zirconium and its alloys: A machine learning potential method

Abstract

Zirconium alloy coating is an important direction for the modification of nuclear cladding materials. Thermal conductivity is a critical property of cladding materials. With extensively studying phonon-electron non-equilibrium energy transfer processes in the thermal transport of zirconium alloy coating, to distinguish the contributions from phonon and electron thermal conductivity of Zr alloys becomes crucial and necessary. In this work, we successfully predicted the lattice thermal conductivities of zirconium, Zr-Sn and Zr-Nb using machine learning potentials. Sn and Nb doping leads to a significant decrease in lattice thermal conductivity, which is mainly due to the alterations in phonon group velocity and phonon scattering. The larger atomic mass of doping elements and weakened interatomic interactions of Zr-Nb together lead to a significant decrease in phonon group velocity. Doping Sn and Nb also increases phonon-phonon scattering rate and three-phonon scattering channels, resulting in a shortening in phonon lifetime and a decrease in lattice thermal conductivity.